Solvent Accessibility Studies of &Bends and
Application to Cyclic Hexapeptides"
K. V. SOMAN and C. RAMAKRISHNAN, Molecular Biophysics
Unit, Indian Institute of Science, Bangalore 560 012, India
Synopsis
The solvent-accessible surface areas (ASAS),of the atoms in tripeptides around the
minimum-energy conformations of the &bend types I, 1', 11, and 11' have been computed
as a first step in the systematic solvent accessibiity study of secondary structures. The
side chains chosen at the two middle positions of the bend are L-Ala, DAla, and Gly.
The ASAs of the hydrogen atoms are reported here and are found useful in determining
the type of P-bends in six examples of cyclic hexapeptides whose crystal structures are
known. Comparison with observation showed that all the @bends in these cyclic hexapeptides were correctly identified by the present method. This points to a possible use
of the method in identifying B-bend types in solution.
INTRODUCTION
Information regarding the exposure of the different atoms, groups,
and residues in a protein to the solvent surrounding it is useful in
understanding its structure and function. The idea of solvent accessibility had been developed, and a quantity known as solvent-accessible
surface area (ASA) had been computed for the first time by F. M.
Richards and coworkers.' The concept of ASA has been applied by
many workers t o the study of protein stability and folding,= as well
as to smaller pep tide^.^^ Recently, Snyder and coworkerslO have computed the exposure of NH hydrogen atoms in a cyclic hexadepsipeptide,
roseotoxin B, by means of ASAs.
The work described here involves the computation of the ASAs of
some P-bend conformations and forms a first step in a systematic study
of the solvent accessibility of the secondary structures in proteins
(helices, &bends, extended strands, and &sheets).
SYSTEMS AND METHODS
The systems used in the study are tripeptides in P-bend conformations of types I, 11, 1', and 1I',I1one of which is shown in Fig. l. The
ideal values of the backbone torsional angles (+2,4J2), (&,4J3) for these
bend types are: type I ( - 60", - 30'7,( - 90",0"); type I1 (- 60",120"), (80",0");
* Part XI1 of the series, Stereochemical Studies on Cyclic Peptides. Part XI is Ramakrishnan, C. & Narasinga Rao, B. N. (1980) Int. J. Pept. Protein Res. 15, 81-85.
Biopolymers, Vol. 24, 1205-1214 (1985)
@ 1985 John Wiley & Sons, Inc.
CCC 0006-3525/85/071205-10$04.00
1206
4
SOMAN AND RAMAKRISHNAN
Fig. 1. Tripeptide in a P-bend conformation showing atom numbering and the
1 hydrogen bond (- - -1.
+
type I' (60",30"),(9o",o"); and type 11' (60", - 1207, (- 80",0").The side
chains L-Ala, DAla, and Gly (abbreviated to L, D, and G, respectively)
are fixed at positions Cg and Cg leading to a total of nine tripeptides,
namely, GL, GD, LG, DG, LL, DD, LD, DL, and GG. For each of these
tripeptides, the potential energy, including hydrogen-bond energy, is
calculated for hydrogen-bonded conformations in the regions of each
of the bend types separately, varying the torsion angles (+2,$2),
a t lo" intervals. Standard potential functions and constants were
used.12 The conformations at the different grid points are then arranged in the increasing order of potential energy. In each case, 100
conformations, starting from the minimum upward, are used for ASA
calculations. In fact, calculations were done only for types I and 11,
since types I' and 11' are, respectively, their mirror images for enantiomeric sequences and have the same potential energy and ASA values.
The method used for computing ASAs is that of Lee and Richards'
and employs the computer programs supplied by the latter. The method consists in rolling a solvent molecule down the surface of the tripeptide, making the maximum permissible contact with each atom.
The van der Waals radii used are1J2J3:C, 1.70 A; C including attached
hydrogens (used for CP,~C;,and the CY of Ala residue), 1.80 A; 0, 1.52
A; N, 1.55 A; H, 1.20 A; and the solvent water molecule, 1.40 A. In
this study, attention has been confined to the ASAs of the hydrogen
atoms alone because they stick out of the polypeptide backbone (as do
the carbonyl oxygens) and, hence, their ASAs will be very sensitive
to changes in conformation. Besides, the possibility of obtaining the
solvent exposure of the hydrogens experimentally makes them the
ideal choice in the efforts to distinguish conformations from ASA values. The ranges over which the ASAs of the hydrogen atoms vary for
these conformations are given in Table I, each section of the table
SOLVENT ACCESSIBILITY OF &BENDS
1207
TABLE I
Ranges of ASAs (in
Az)of Hydrogen Atoms for the Different Types of P-Bends with
Different Middle Residues
Type of P-Bend
Atom
Gly-DAla bend
Hz
H3
H4
Hr;l
Hq2
Hf
L-Ala-Gly bend
Hz
H3
H4
H?
Hg'
Hf2
DAla-Gly bend
Hz
H3
H4
Hr;
H3'
Hf2
I
I'
I1
11'
18.1-20.2
2.8-10.4
0.0-5.2
22.1-24.6
18.7-21.4
12.9-19.8
17.5-20.9
5.2-15.3
0.0-5.1
18.5-21.6
22.1-24.6
19.8-22.1
20.1-21.6
4.618.5
0.0-5.6
22.2-24.5
14.9-19.7
18.3-21.9
20.3-21.6
3.3-12.7
0.0-4.3
15.9-19.7
22.2-24.1
11.G19.7
17.5-20.9
5.2-15.3
0.0-5.1
22.1-24.6
18.5-21.6
19.S22.1
18.1-20.2
2.8-10.4
0.0-5.2
18.7-21.4
22.1-24.6
12.9-19.8
20.3-21.6
3.3-12.7
0.0-4.3
22.2-24.1
15.9-19.7
11.6-19.7
20.1-21.6
4.418.5
0.0-5.6
14.9-19.7
22.2-24.5
18.3-21.9
13.0-14.7
5.3-12.0
0.0-5.2
16.3-18.6
22.6-25.0
15.5-22.6
18.5-20.7
5.5-15.3
0.0-5.2
19.4-21.9
15.5-22.6
22.6-24.6
14.8-15.7
8.8-16.9
0.0-5.6
12.7-16.1
15.3-21.9
22.5-24.8
20.7-21.4
1.5-7.4
0.0-5.7
18.9-20.1
20.7-26.1
13.9-22.2
18.5-20.7
5.5-15.3
0.0-5.2
19.4-21.9
22.6-24.6
15.5-22.6
13.0-14.7
5.3-12.0
0.0-5.2
16.3-18.6
15.5-22.6
22.6-25.0
20.7-21.4
1.5-7.4
0.0-5.7
18.9-20.1
13.9-22.2
20.7-26.1
14.S15.7
8.8-16.9
0.0-5.6
12.7-16.1
22.524.8
15.3-21.9
13.0-14.7
2.G7.6
0.0-5.2
16.3-18.6
12.8-19.7
18.1-20.7
5.2-15.3
0.0-5.1
19.621.9
19.s22.1
14.7-15.7
4.7-17.6
0.0-5.6
12.5-16.9
18.3-21.9
20.7-21.4
0.3-3.5
0.M.4
18.6-20.0
11.3-19.4
18.1-20.7
5.2-15.3
0.0-5.1
19.4-21.9
19.8-22.1
13.0-14.7
2.G7.6
0.0-5.2
16.3-18.6
12.S19.7
20.7-21.4
0.3-3.5
0.0-4.4
18.6-20.0
11.3-19.4
14.7-15.7
4.7-17.6
0.0-5.6
12.5-16.9
18.3-21.9
(continued)
L-Ala-L-Ala bend
H
Z
H3
H4
H8
Hf
~ A l a - n A l abend
H2
H3
H4
H5
H?
1208
SOMAN AND RAMAKRISHNAN
TABLE I icontznuedl
Type of P-Bend
Atom
I
I’
I1
11’
12.7-15.3
5.1-12.2
14.8-15.7
5.1-12.5
O.M.3
12.7-16.1
12.619.5
20.7-21.4
1.143.1
0.C5.7
18.6-20.6
18.3-22 .O
L-Ala-BAla bend
Hz
H4
0.0-5.1
H5
H3
16.1-18.9
19.9-22.1
18.5-20.6
3.610.5
0.C5.2
19.P21.9
12.8-19.8
18.5-20.6
3.4-10.5
0.C5.2
19.621.9
12.8-19.8
12.7-15.3
5.1-12.2
0.0-5.1
16.1-18.9
19.9-22.1
20.7-21.4
1.143.1
0.0-5.7
18.620.6
18.3-22.0
14.8-15.7
5.1-12.5
0.0-4.3
12.7-16.1
12.6-19.5
18.1-20.2
5.3-15.3
0.C5.2
22.1-24.6
18.7-21.4
22.6-25.0
15.5-22.6
18.1-20.2
5.S15.3
0.C5.2
18.7-21.4
22.1-24.6
15.5-22.6
22.625.0
20.3-21.6
7.6-17.6
0.0-5.6
22.3-24.2
15.8-19.7
15.3-22.4
22.0-24.8
20.3-21.6
7.6-17.6
0.0-5.6
15.&19.7
22.3-24.2
22.0-24.8
15.3-22.4
H3
D-Ala-L-Ala bend
H2
H3
H4
m
H3
Gly-Gly bend
H2
H,
H4
Ha’
Hp
Hg’
Hf2
giving the values corresponding to one of the nine tripeptides. The
values in Table I reveal the following points relevant to distinguishing
bend types.
1. The ASA of H, (see Fig. 1 for atom numbering) is uniformly low
(5.7 Az or lower) as this is directly involved in the 4
1 hydrogen
bond. Thus, the ASA of H, cannot be expected to help in distinguishing
bend types.
2. The H, atom is well exposed to the solvent, as may be seen from
the table, although its ASA values vary within a narrow range of
about 2 A2.
3. The ASAs of the H3 atom vary a good deal. The ranges for the
different bend types are different, although overlap is common. The
wider spectrum of the ASA of H3 is quite in accord with our expectations, because this atom forms part of the middle peptide unit of the
&bend, whose tilt accounts for the difference between the type I and
type I1 bends.
4. The total number of Ha atoms varies from 2 to 4, depending on
the residues of Cq and Cg (Fig. 1).Their ASAs are generally high, vary
from one bend type to another, and the ranges are different despite
overlap.
5. Considering the standard bend varieties L-L type I, L-D type 11,
+
SOLVENT ACCESSIBILITY OF &BENDS
1209
D-D type 1’,and D-L type 11’[LL(I), LD(II), DD(I’), DL(II’)],the H, atom
is more accessible in LD(I1) than in LL(I), whereas for Hg the reverse
is the case. In the case of H3, the ASA ranges for LL(1) and LD(I1)
overlap somewhat, with the latter being more accessible. DD(1’) and
DL(I1’) are mirror images of LL(1) and LD(II), respectively.
APPLICATION OF ASA CALCULATIONS TO
CYCLIC HEXAPEPTIDES
Procedure
The indication that bend types can be distinguished using hydrogen
ASA was checked by application to cyclic hexapeptides (CHP). From
among the dozen or so CHPs whose x-ray crystal structures are available, the six listed in Table I1 were chosen for our study.
The usefulness of the procedure is based on the assumption that it
will be possible in the not-too-distant future to obtain experimentally
the ASA values of hydrogen atoms of peptides in solution from nmr
spectroscopy. However, lacking experimental data, it is necessary at
present to simulate the “observed” values of ASAs by computation
from the x-ray structure coordinates. The following procedure is used
to identify the bend type:
1.Where the positions of hydrogen atoms are not reported, they are
fixed using standard geometry.
2. All non-Gly side chains are stripped of their side-chain atoms
from Cu onward in order to equate them to L-Ala or D-Ala. The ASAs
of the hydrogen atoms in these simplified molecules are calculated as
in the case of &bends, and the values so obtained serve as “observed”
values.
3. The bend type is now determined by comparing the “observed”
ASAs with the theoretical values for the bends given in Table I. If
TABLE 11
Cyclic Hexapeptides Used in this Study
Name
CHPl
CHPP
CHP3
CHP4
CHP5
CHP6
a F r o m Ref.
bFrom Ref.
From Ref.
From Ref.
<’ From Ref.
14.
15.
16.
17.
18.
Sequence
1210
SOMAN AND RAMAKRISHNAN
di, dil, . . . are the differences between the “observed” ASA of the j t h
hydrogen atom (in each bend) and the mean of the lower and upper
limits given in Table I for each bend type, then the sum of the differences is given by
n
D,=
c ldjl
j=1
where j = 1,2, . . . , n corresponds to the n hydrogen atoms in the
peptide and t stands for the bend type. The bend type for which the
value of D is the lowest has been taken as the most possible bend.
RESULTS
The details of the steps in arriving at the bend types from the ASA
values are first illustrated with one cyclic hexapeptide cycZo(L-Ala-LAla-Gly-Gly-L-Ala-Gly)(CHPl), and the results on the other molecules
are presented in summary form.
The ASAs of the hydrogens in CHPl are shown schematically in
Fig. 2. The first step is to determine the location of the bends in the
CHP. In the diagram, it can be seen that the atoms H, and H, have
very low ASAs (0 and 1.4 Az)clearly showing that these are involved
in hydrogen bonding and that the two bends are composed of residues
6-1-2-3 and 3-4-5-6.
The next step lies in identifying the type of bend, the details of
)la
H’
,
HJ
G:c(22)
HA (8.7)
ti”
(8.7) (8.7)
(20.5)
Fig. 2. Schematic diagram of the cyclic hexapeptide cyclo(L-Ala-~Ala-Gly-Gly-L-AlaGly). The values in parentheses are the ASAs of the hydrogen atoms in the molecule.
SOLVENT ACCESSIBILITY OF &BENDS
1211
TABLE I11
ASA Values of the Hydrogen Atoms of cyclo(L-Ala-L-Ala-Gly-Gly-LAla-Gly)and
Comparison with Corresponding Values for Typical Bendsa
~~
ASA of Hydrogen Atoms
Sum of
Differences,
(A2)
D,
Bend Type
Hz
H3
Hg’
HP
HQ
(152,
Bend 1: L-Ala-L-Ala
“Observed”
I
I1
I’
11‘
11.70
13.85
(2.15)
15.20
(3.50)
19.40
(7.70)
21.05
(9.35)
14.80
17.45
(2.65)
14.70
(0.10)
20.65
(5.85)
19.30
(4.50)
5.00
5.10
(0.10)
11.15
(6.15)
10.25
(5.25)
1.90
(3.10)
12.00
16.25
(4.25)
20.10
(8.10)
20.95
(8.95)
15.35
(3.35)
9.15b
17.85
27.75
20.30
Bend 2: Gly-L-Ala
“Observed”
I
I1
I’
11’
a
14.20
19.15
(4.951
20.85
(6.65)
19.20
(5.00)
20.95
(6.75)
8.70
6.60
(2.10)
11.45
(2.75)
10.25
(1.55)
8.00
(0.70)
8.70
23.40
(14.70)
23.35
(14.65)
20.05
(11.351
17.80
(9.10)
20.50
20.10
(0.40)
17.30
(3.20)
23.35
(2.85)
23.15
(2.65)
8.70
16.35
(7.651
20.10
(11.40)
20.95
(12.25)
15.65
(6.95)
29.80
38.65
33.00
26.15b
Differences between the two are given in parentheses.
value.
Bend type with the lowest 0,
which are given in Table 111. The first line gives the “observed” ASA
values for each hydrogen atom, and below it are given the corresponding ideal values for the bend types obtained from the appropriate
section of Table I. The difference (Dj)between the “observed” and the
ideal values is given in parentheses. From the sum of the D values in
the last column, it can be seen that the first bend belongs to type I
and the second to type 11‘,although in the latter case, the difference
between the lowest and the next higher ASA values is not as pronounced as in the former. These deductions agree with those made
from the (+,$I values.
The complete results on the six CHPs are given in Table IV, where
the “observed” (simulated) hydrogen ASA values, along with the assignment of the bend types based on them (corresponding to the lowest
D value), are listed. These assignments are seen to agree, in all cases,
with those based on the (+,+) values given in the last four columns.
CHPl
1-2
4-5
CHP2
6-1
3-4
CHP3
1-2
4-5
CHP4
4-5
1-2
CHP5
2-3
5-6
CHP6
1-2
4-5
No.
-
4.0
10.0
6.5
13.0
9.8
11.3
6.6
6.6
9.3
15.7
3.1
19.8
12.6
13.2
12.7
12.7
L-Ala-L-Ala
L-Ala-L-Tyr
D-Leu-L-Leu
DLeu-L-Leu
L-Leu-L-Phe
BLeu-BPhe
Gly-L-His
L- Ala-Gly
19.77
13.5
9.0
5.5
17.6
11.4
11.8
11.8
17.5
-
-
-
-
13.1
13.2
-
19.8
24.7
-
16.2
14.9
15.6
9.8
24.4
14.62
-
12.0
8.7
13.9
19.0
I
16.24
13.4
15.30
-
-
(A2)
14.8
20.5
17.5
12.6
13.6
15.7
-
-
8.7
Gly-Gly
D-Ala-DAla
-
Hg’
5.0
8.7
H3
11.7
14.2
H2
“Observed” ASA Values
L-Ala-1,-Ala
Gly-L-Ala
Name
Residues in Bends
I
I’
11’
11’
I
11’
- 59
59
77
66
-56
61
54
- 62
- 70
66
I
I’
I
I’
84
- 53
32
38
32
- 32
-116
-121
-135
- 35
-
15
- 15
106
131
106
85
115
115
-
- 87
-95
-77
91
- 95
-
-
- 84
(4,+) Values
- 43
-113
Observed
I
11’
Bend
TABLE IV
ASAs of the Hydrogen Atoms in Six Cyclic Hexapeptides and Assignments of P-Bend Types
(deg)
36
-36
-4
-4
-10
-10
13
-6
16
-31
Z
z
k
g
x
!Jj
+
z
u
k
-9O ! 2
SOLVENT ACCESSIBILITY OF P-BENDS
1213
DISCUSSION AND CONCLUSION
In the present paper, only the ASAs of the hydrogen atoms in bends
have been studied. This limited study points to the usefulness of ASAs
in understanding and identifying some conformational features in
small peptides. The described method for distinguishing bend types in
cyclic hexapeptides is an offshoot of the calculations on ideal bend
types. The agreement between the simulated “observed” ASA values
of hydrogens and those calculated for the ideal bend types shows that
if the ASA values for the CHPs in solution could be obtained experimentally, it would be possible t o predict the type of P-bend occurring
in them using the procedure presented here.
In the present study, we have used simulated values of proton ASAs.
However, there are indications that such values may soon become
experimentally available. For example, nitrosyl-induced enhancement
in T I relaxation rates of protons have been used as a probe of conformation in peptide~.~gpz~
Since the phenomenon is correlated to the extent of exposure of the proton to the surrounding solvent, it must be
possible to derive ASA values from the enhancement in T I values.
These can then be used to distinguish bend types.
Side-chain atoms beyond CY have not been included in the calculations of ASA values described here. However, it would be useful to
have a qualitative idea as to how a larger side chain can shield the
hydrogen atoms from the solvent. A side chain at Cq reduces the ASA
of the H, atom considerably (by 5 Az or more, depending on the bulkiness of the group), except when the torsion angle,
is around 180”,
at which value the side chain points away from H,. The ASA of the
Hp atom would be affected to a smaller extent, whereas H3 and Hg
ASAs would hardly be affected. Similarly, a longer side chain at Cg
would cut off parts of H, and Hg from solvent, but this effect may be
masked by the larger spectrum of variation of the ASAs of these two
atoms. A side chain at Cg has hardly any effect on the ASAs of H,
and H;.
The calculations on bends described here form part of our ASA study
of secondary structures. The success of the present application is an
encouraging sign to calculate ASAs for a general bend, defined in our
earlier
which include those that do not have 4
1 hydrogen
bonds. It is possible to extend the computation to the different types
of helices, extended strands, and &sheets and to compare them with
known protein structures so that information regarding the tendencies, if any, of the different elements of secondary structures to bury
or expose themselves during the process of protein folding can be
derived. Some of these studies are under way.
x’,
+
We would like to thank Professor Kenneth D. Kopple for useful suggestions and
discussions. The solvent-accessibility programs were supplied by Professor F. M. Richards.
1214
SOMAN AND RAMAKRISHNAN
References
1. Lee, B. & Richards, F. M. (1971) J. Mol. Biol. 55, 379400.
2. Chothia, C. (1974) Nature 248, 33S339.
3. Chothia, C. (1975) Nature 254, 304-308.
4. Chothia, C. (1976) J. Mol. Biol. 105, 1-12.
5. Chothia, C. & Janin, J . (1975) Nature 256, 705-708.
6. Ponnuswami, P. K. & Manavalan, P. (1976) J. Theor. Biol. 60,481-486.
7. Manavalan, P. & Ponnuswami, P. K. (1977) Biochem. J. 167, 171-182.
8. Genest, M., Vovelle, F., Ptak, M., Maigret, B. & Premilat, S. (1980) J. Theor. Biol.
87, 71-84.
9. Vovelle, F., Genest, M., Ptak, M., Maigret, B. & Prelimat, S. (1980) J. Theor. Biol.
87, 85-95.
10. Snyder, J. P. (1984) J. Am. Chem. Soc. 106, 2393-2400.
11. Venkatachalam, C. M. (1968) Biopolymers 6, 1425-1436.
12. Ramachandran, G. N. & Sasisekharan, V. (1968) Adu. Protein Chem. 23, 283-437.
13. Bondi, A. (1964) J. Phys. Chem. 68, 441451.
14. Hossain, M. B. & van der Helm, D. (1978) J. Am. Chem. Soc. 100, 5191-5198.
15. Karle, I. L., Gibson, J . W. & Karle, J. (1970) J. Am. Chem. Soc. 92, 3755-3760.
16. Yang, C-H., Brown, J. N. & Kopple, K. D. (1981) J. Am. Chem. Soc. 103, 17151719.
17. Varughese, K . I., Kartha, G. & Kopple, K. D. (1981) J. Am. Chem. Soc. 103,331C3313.
18. Hossain, M. B. & van der Helm, D. (1979) Acta Crystallogr.,Sect. B 35, 2634-2638.
19. Niccolai, N., Valensin, G., Rossi, C. & Gibbons, W. A. (1982) J. Am. Chem. Soc.
104, 1534-1537.
20. Kopple, K. D. (1983) Znt. J. Pept. Protein Res. 21, 43-48.
21. Ramakrishnan, C. & Soman, K. V. (1982) Znt. J. Pept. Protein Res. 20, 21S237.
Received May 8, 1984
Accepted November 27, 1984